THE FOUR GOSPELS - A Study of Origins, The Manuscript Tradition, Sources, Authorship, & Dates. By B.H. Streeter, Hon.D.D.Edin. Fellow of Queen's College, Oxford; Canon of Hereford. First published by MacMillan & Co Limited 1924. - Prepared for katapi by Paul Ingram 2004.

Chapter IX


Home | Diagram – the synoptic sources | synopsis | unconscious assumptions | Jerusalem, Caesarea & Antioch | 2 recensions of Q | parallel versions | overlapping of resources | Matthew's method of conflation | the Sermon on the Mount | Judaistic tendency of M | overlapping of Mark & M | the great discourses of Matthew | infancy narratives | conclusion | list of MSS | (pages 223-270)


unconscious assumptions

Three unconscious assumptions, which have led to a misinterpretation of the available evidence.

  1. The name TWO DOCUMENT HYPOTHESIS suggests that no other sources used by Matthew and Luke are comparable to the "Big Two." Hence an undue importance has been assigned to Q as compared with the sources used by Matthew or Luke only.
  2. It is assumed that a hypothesis that reduces the number of sources to a minimum is more scientific.
  3. It is taken for granted that the same saying is not likely to have been reported by more than one independent authority.
    But a plurality of sources is historically more probable. In particular, if Mark is the old Roman Gospel, it is antecedently to be expected that the other Gospels conserve the specific traditions of Jerusalem, Caesarea, and Antioch.

Jerusalem, Caesarea, and Antioch

A priori probabilities in regard to the traditions of these Churches.
The non-Marcan matter in Luke has been analysed further into at least two sources, Q and L; similarly we may expect to find that Matthew used a peculiar source, which we may style M, as well as Q.
The Judaistic character of much of the material in M suggests a Jerusalem origin.
L has already been assigned to Caesarea.
Q may be connected with Antioch.
Most probably Q is an Antiochene translation of a document originally composed in Aramaic perhaps by the Apostle Matthew for Galilean Christians.
On this view our first Gospel is a combination of the traditions of Jerusalem, Antioch, and Rome, while the third Gospel represents Caesarea, Antioch, and Rome. The fact that the Antiochene and Roman sources are reproduced by both Matthew and Luke is due to the importance of the Churches; it is no evidence that the other sources are less authentic.
Although, however, historical considerations favour a Four Document Hypothesis, the verification of the hypothesis must depend entirely on the results of a critical study of the documents apart from any theory as to the geographical affiliation of any particular source.

the theory of two recensions of Q

It has already been recognised by critics that the Two Document Hypothesis in its simplest form has broken down. The theory of two recensions of Q, designated as QMt. and QLk., has been put forward to meet the difficulty. But on examination this theory is seen, not to solve, but to disguise, the problem by an ambiguity latent in the symbols used.

parallel versions

General considerations as to the extent and ways in which collections of sayings or parables, though made independently, would nevertheless inevitably overlap.
Evidence from non-Canonical sources as to the existence of independent parallel versions of sayings of Christ.

the overlapping of sources

Three clear cases of this in the Gospels.
(1) Mark and Q.
(2) Mark and L.
(3) The collections of Parables in Matthew and Luke.
Evidence that the versions of the Lost Sheep found in Matthew and Luke were drawn from two different sources; a fortiori this holds good of the Marriage Feast and the Talents.
All analogies, then, suggest that Q and M would overlap.

Matthew's method of conflation

The meticulous way in which Matthew conflates his sources illustrated by a study of two examples where he is combining Mark and Q.
This compels us to formulate a new principle of Synoptic criticism: "Wherever parallel passages of Matthew and Luke show substantial divergence, editorial modification is a less probable explanation than conflation by Matthew of the language of Q with that of some other version."

the sermon on the mount

In view of the evidence as to overlapping of sources and Matthew's method of conflation set out in the two preceding sections, we may now test the hypothesis that Q and M overlapped.
The Sermon on the Mount is a conflation of a discourse in Q (approximately represented by Luke's Sermon on the Plain) and a discourse from M which happened to begin with a series of Beatitudes very different in detail from those in Q and to contain a divergent version of "Love your Enemies" and a few other sayings. Into this conflated discourse Matthew has introduced some additional fragments of Q, which Luke gives in his Central Section, presumably in their original context.
The Woes to the Pharisees, Mt.xi., is probably another case of conflation by Matthew of discourses in Q and M which had certain points of contact Luke's version again being nearer to Q.

Judaistic tendency of M

The question must be faced, did the Judaistic sayings in Matthew stand originally in Q, being omitted by Luke owing to his pro-Gentile proclivities, or are they to be assigned to M?
Reasons for choosing the latter alternative.

overlapping of Mark and M

Three passages in Matthew, in the main clearly derived from Mark, contain certain added details of a specially interesting character. Possibility that these were derived from a parallel version in M.

the great discourses of Matthew

Evidence that the five great discourses of Matthew are agglomera­tions by the editor of the Gospel, and do not correspond to collocations of the material in an older source.

the infancy narratives

The first two chapters of Matthew are probably derived from oral sources, but the corresponding section in Luke is more likely to have been found by him in a written document, possibly Hebrew.
Some points of textual criticism bearing on the evidence for the Virgin Birth.


The Four Document Hypothesis, besides explaining a number of facts, which are not accounted for by the Two Document Hypothesis, materially broadens the basis of evidence for the authentic teaching of Christ.
N.B.—The Diagram: The Synoptics and their Sources (p. 150 above) should be referred to in connection with this chapter.



unconscious assumptions

THE psychologists are all warning us of the peril of the UNCONSCIOUS MOTIVE. It is against "unconscious assumptions" that critics of the Gospels most need to be on their guard.

  1. It is unfortunate that the name TWO DOCUMENT HYPOTHESIS should have been given to the theory that the authors of the First and Third Gospels made use of Mark and Q, for it conceals the unconscious assumption that they used no other documents, or, at least, none of anything like the same value as the "Big Two." Hence a quite illusory pre-eminence has been ascribed to the document Q in comparison with the sources for our Lord's teaching made use of by Matthew or by Luke alone. To this illusion I must confess that I have been myself for many years a victim. The idea has grown up that it is just a little discreditable to any saying of our Lord if it cannot be traced to Q. Immense efforts are accordingly made to extend the boundaries of Q as much as possible as if a sentence of exclusion from this document meant branding the excluded saying with a reputation of doubtful historicity. Much of what is clearly authentic teaching of Christ quite half of the Sermon on the Mount, for instance is found in only one Gospel. An effort, then, must be made to get all this material somehow or other assigned to Q; and ingenious motives must be discovered to explain why the other evangelists omitted it. Once, however, the "unconscious assumption" of some special superiority of Q is brought up into the daylight of clear consciousness, a moment's consideration will show that it is wholly baseless. One has only to mention the fact that hardly any of the parables are found in Q to realise that a large part of the most obviously genuine, original, and characteristic teaching of our Lord is derived, not from Q, but from sources peculiar to Matthew or Luke.
    The Good Samaritan,
    the Prodigal Son,
    the Pharisee and the Publican,
    are peculiar to Luke;
    the Labourers in the Vineyard,
    the Pearl of Great Price,
    are given by Matthew alone.
    There cannot be the slightest presumption that a source which lacked such material as this is a more reliable authority than those which contained it.
    Some scholars, indeed, have been so far hypnotised by the prestige of Q that, from the possible absence from Q of the longer narrative parables, they have drawn the conclusion that such parables formed no part of the original teaching of our Lord but are developments in later tradition, though probably in some cases being expansions of shorter authentic sayings. Nothing could be more absurd. Our Lord was above all a popular teacher; it was the common people who heard Him gladly. But everybody knows that a story told vividly and in detail is the one thing most likely to attract the attention and to remain in the memory of a popular audience. A friend once said to me, "You can preach the same sermon as often as you like, provided you don't repeat your illustrations; but tell the same story twice, and, even if the rest of your sermon is on a totally different topic, people will say that you repeat yourself."
    If one considers the teaching of Christ from the standpoint of the psychology of everyday life and not of academic theory, it is obvious that the parables, and that in their most graphic and least curtailed form, such as we find in Luke, are just the element most likely to belong to the earliest stratum of tradition. Why the author of Q included so few (or, possibly, none at all) of them, we cannot say, any more than we can say for certain why he did not include an account of the Passion. Probably the reason for both omissions was the same. He wrote to supplement, not to supersede, a living oral tradition. Both the longer parables and the Passion story were easy to remember, and every one knew them; and what he was most concerned to write down was something which was either less well known or easier to forget.
  2. Another equally misleading assumption, again more or less unconscious, has been the idea that antecedent probability is in favour of a hypothesis which so far as possible reduces the number of sources used by Matthew and Luke, and minimises the extent and importance of the sources of material peculiar to one Gospel. This is due to a confusion of thought. Since Matthew and Luke appear to have written in churches in every way far removed from one another, that hypothesis is the most plausible which postulates the smallest number of sources used by them in common. But that same removedness only increases the likelihood that each had access to sources not known to the other, outside the two (Mark and Q) that they concur in using.
  3. There is yet a third false assumption current that it is improbable that the same or similar incidents or sayings should have been recorded in more than one source, and that, therefore, if the versions given by Matthew and Luke respectively of any item differ considerably, this is to be attributed to editorial modification. On the contrary, given the existence of independent reports of the sayings of a great teacher, these would inevitably overlap and would sometimes give almost identical, at other times widely different, versions of the same saying. The overlooking of this consideration has had fatal effects on Synoptic criticism.

One reason why these erroneous assumptions have held sway so long is that the Synoptic Problem has been studied merely as a problem of literary criticism apart from a consideration of the historical conditions under which the Gospels were produced. A factor of cardinal importance has been ignored the preponderating influence of the great Churches in the deter­mination of the thought and literature of primitive Christianity. Mark, says an ancient and probably authentic tradition, was written in Rome a long way from Palestine. But Jerusalem and Caesarea, the two great Palestinian Churches, and Antioch, the original headquarters of the Gentile Mission, must each have had a cycle of tradition of its own. It is in the last degree improbable that the characteristic traditions of any of these three Churches have completely disappeared. It is far more likely that, in one form or another, they are incorporated in the Gospels that were ultimately accepted as exclusively canonical by the Church at large. Thus traces of at least three different cycles of tradition, besides the material derived from Mark, are what antecedently we should expect a critical examination of Matthew and Luke to reveal.

Jerusalem, Caesarea, and Antioch

Accordingly, before entering upon a critical examination of actual documents, it will be worthwhile to consider the historical probabilities in regard to the collection and transmission of sayings of our Lord by Christians of the first generation.

In Jerusalem it is on the whole likely that the sayings would for some considerable time be handed down in oral tradition after the manner of the sayings of the Rabbis, and that in the original Aramaic. But in the Greek-speaking Churches a beginning would be made at writing them down almost at once. Collections of sayings regarded as especially valuable for the instruction of converts would very soon be formed in various Churches. But the Churches of Antioch and Caesarea are those where we should expect to find not only the earliest, but also the most considerable and the most valuable, collections written in the Greek language. For these were the first Gentile Churches to be founded, and also, from their geographical position, were peculiarly well situated for procuring authentic material. Peter himself had indeed visited both these Churches at a very early date. But sooner or later possibly not till the flight of Christians to Pella just before the final siege the Jerusalem collection also would be committed to writing. Once that was done, it would sooner or later reach Antioch or Caesarea, and a Greek translation of it would be made and so become available to the Gentile Churches. The antecedent probabilities, then, are that there would be at least three considerable collections of the teaching of Christ, associated with the Churches of Jerusalem, Antioch, and Caesarea. The Church of Jerusalem was for a time "knocked out" by the Jewish War. But Antioch and Caesarea were sufficiently influential to secure that the traditions, which they specially valued, did not completely disappear. They were also Churches to which those who wished for authentic information about our Lord could readily resort.

Supporters of the Two Document Hypothesis usually assign to Q the bulk of the discourse material in Matthew, apart from the longer parables. This involves as a corollary either that Luke has made very drastic omissions from Q or else that Matthew used an expanded edition QMt. a hypothesis to which I adduce objections below, p. 235 ff. But in view of the a priori probability of there being three cycles of tradition available, besides that of Mark, I would suggest the simple hypothesis that, just as Luke ultimately goes back to at least two sources besides Mark, viz. Q and L, so it is with Matthew. Provisionally I will assign to Matthew's third source all discourse peculiar to Matthew, and also that part of the material usually assigned to Q which differs so much from its Lucan parallels as to have suggested the need for the QMt. hypothesis; retaining Q as the name of the source of the close parallels only.
This third source of Matthew it will be convenient to call M.

The material peculiar to Matthew, in sharp contrast to Luke's, is characterised by a conspicuously Jewish atmosphere; and, though rich in anti-Pharisaic polemic, it asserts the obligation of obeying not only the Law but "the tradition of the scribes," and it has a distinctly anti-Gentile bias. It reflects the spirit and outlook with which in the New Testament the name of James is associated; though James himself, like most leaders, was doubtless far less extreme than his professed followers. The source M will naturally be connected with Jerusalem, the head­quarters of the James party.

The Caesarean tradition, we naturally surmise, has survived by its incorporation in the Third Gospel. The reasons for connecting Luke's special source L with Caesarea have been given above (p. 218 f.) and need not be repeated.

Antioch, then, remains as a possible place for the origin of Q. In the Oxford Studies I suggested a Palestinian origin, with the probability that the Apostle Matthew was its author. [I also suggested that Mark was written for a Church that already possessed a collection of Christ's sayings and desired a biography to supplement it. The suggestion, I still think, is worth consideration, although the conclusion there drawn that this collection must have been Q was too hasty. Why should it not have been the local Roman collection of sayings from which Clement seems to quote? Cf. p. 240 n.] The two suggestions are not in the slightest degree incompatible. The source Q, with which the student of the Synoptic Problem is concerned, is a Greek document which the authors of the First and Third Gospels had in common, and the fact that this Greek document was known to the authors of both these Gospels means that it probably came to them with the backing of the Church of some important Greek city. But that is no reason why it may not have been a translation of an Aramaic work by Matthew possibly with some amplification from local tradition. What became of the Twelve Apostles is one of the mysteries of history. The resident head of the Jerusalem Church was, not one of the Twelve, but James the brother of the Lord. From Galatians and Acts we should gather that to Peter and the sons of Zebedee Jerusalem was for a time a kind of headquarters. But in regard to the rest there is no tradition, which, either from its early date or its intrinsic probability, deserves credence. But we know from the Rabbis that for many centuries Capernaum was a great centre for "Minim" or Christians, so that it is probable that others of the Twelve made that city their headquarters. Geographically Capernaum is between Antioch and Jerusalem, and some Christian trader from Antioch having business at Capernaum, or in some city of Decapolis, may well have come across a collection of sayings made by Matthew and brought it home.

The hypothesis that Q emanated from the (perhaps, freer) atmosphere of Galilee and became the primitive "gospel" of a Gentile Church, like Antioch, accounts for its inclusion of the saying (Lk.xvi.16, Mt.xi.13),

The law and the prophets were until John.

It also explains at once the puzzling fact that in a document, otherwise apparently entirely confined to discourse, there should have stood the one single narrative of the Centurion's Servant. That story leads up to, and gives the facts that called forth from Christ, the saying,

I have not found so great faith, no, not in Israel.

At a time when the Judaising section of the Church wished to give the uncircumcised an inferior status, that story was in itself a charter of Gentile liberty.

The Greek translation of Q, at any rate, must have been made for the use of a Greek Church, and since, if we regard the material peculiar to Luke as representing the tradition of Caesarea, that city is ruled out, Antioch, the first capital of Gentile Christianity, is the most likely place of origin. In Ch. XVII. I shall give reasons for supposing that our Gospel of Matthew was written in Antioch. There is also, for what it is worth, a tradition, found in Eusebius and in the Latin Prologues to the Gospels, which has some support in the occurrence in D, etc., of a WE SECTION, Acts xi.27, that Luke was by descent a Syrian of Antioch. I should not care to lay much weight on either of these considerations as evidence for connecting Q with Antioch, but so far as they go they are in favour of the connection.

If the suggestions put forward above be accepted, it would follow that Matthew's Gospel represents a combination of the primitive "gospels" of Jerusalem, Antioch, and Rome; while Luke's is ultimately based on those of Caesarea, Antioch, and Rome. Either of these combinations would be eminently reasonable from the point of view of the authors of the Gospels, who would naturally set the highest value on sources so weightily authenticated. The hypothesis would also explain both why these Gospels seem never to have any serious competitors in the Church, and why so little of authentic tradition survives outside the canonical Four. Since the specifically Ephesian tradition may be supposed to be reproduced in John, there did not exist anywhere any considerable body of tradition authenticated by any important Church, which was not represented in one of the Four Gospels.

But if this view is correct it means that the Roman and the Antiochene sources are made use of twice over. In view of the prestige and wide influence of these two Churches this is not surprising. But the historian must realise that the fact that Mark and Q were used by the editors of two later Gospels does not create any presumption that, because a thing occurs in Mark or in Q, the historical evidence for it is twice as strong as if it occurred once only either in the Jerusalem or in the Caesarean tradition. And this last consideration, I would observe, is not substantially affected, supposing that the scheme of connection between the several sources and particular churches, which I have suggested, is not exactly correct. All that I wish to press is the broad principle that a plurality of sources is antecedently probable, and the fact that the relative historical value of a source is not increased by the number of times it is copied.

I suggest, then, that we should, provisionally at any rate, abandon, not the theory that Matthew and Luke made use of Mark and Q, but the conception of a "Two Document Hypothesis"; and explore the possibilities of substituting for it that FOUR DOCUMENT HYPOTHESIS which from the stand­point of historical probability seems to have far more to recommend it.

A Four Document Hypothesis has claims to investigation quite apart from the theory as to the geographical affiliations of particular sources suggested above. To avoid, therefore, complicating critical with geographical questions, I shall for the rest of this chapter use the symbol M for the source of the discourse and parables peculiar to Matthew; but, for reasons that will be obvious to any student of the Synoptic Problem, I shall not use it to include any narrative peculiar to Matthew. M and Q will be of much the same length. There are eight parables ( = 59 verses) peculiar to Matthew not including the Lost Sheep, the Marriage Feast, and the Talents (= 34 verses) to which must be added approximately 140 verses of discourse of the same character as the bulk of Q. The passages, which Hawkins reckons as probably belonging to Q, (he includes the parables of the Lost Sheep and Talents but not the Marriage Feast) total a little less than 200 verses. Thus, if we assign the bulk of the discourse and parables peculiar to Matthew to M, we have a document quite as considerable in extent as Q. This, however, is merely a matter of arithmetic; the points on which our argument will turn are:

  1. the evidence that M and Q to some extent overlapped;
  2. the Judaistic character of the source M.


the theory of two recensions of Q

The "Two Document Hypothesis," so far as it concerns the non-Marcan element in Matthew and Luke, has broken down. But the breakdown has been concealed by the hypothesis that Matthew and Luke made use of two different recensions of Q, which have been styled respectively QMt. and QLk.. The most thoroughgoing and scientific attempt to work out this distinction in detail which has come into my hands is by an American scholar, Mr. C. S. Patton [Sources of the Synoptic Gospels, The Macmillan Company, New York, 1915.]. The idea of two recensions of Q is at first sight attractive; but the moment one attempts to visualise to one's mind's eye the exact kind of documents implied by the symbols QMt. and QLk., its attractiveness begins to wane. The symbol Q, by itself, stands for a perfectly definite concept a written document from which both Matthew and Luke made copious extracts with some slight amount of editorial change. Also since Matthew and Luke each omit some passages from Mark which the other retains, and may be presumed to have treated Q in the same way it is legitimate to suppose that Q contained certain passages that occur only in one Gospel, so long as we recollect that the identification of these can never be more than a "skilful guess."
But to what definite concept do the symbols QMt. or QLk. correspond?

These symbols are intended to imply two things:

  1. That the document Q did not reach the authors of the First and Third Gospels in its original form, but with extensive interpolations the interpolations in Matthew's copy being quite different from those in Luke's. We note, however, that each of these sets of interpolations is as considerable in extent as the original Q to which they are supposed to have been added. And since the additions in Matthew's copy are quite different from those in Luke's, their respective additions must have been derived from two totally different cycles of tradition. It follows that at least one, and probably both, of these cycles must have emanated from a locality or informant different from that of the cycle embodied in the original Q. We are forced, then, to assume the existence of at least one, and probably two, cycles of tradition besides Q. But, if so, what presumption is there that the material preserved in these other cycles reached Matthew and Luke attached to Q and not in independent sources?
  2. The symbols QMt. and QLk. are mainly intended to meet the difficulty that some of the parallels between Matthew and Luke the two versions of the Beatitudes, for instance are so inexact that it is not possible to suppose that Matthew and Luke derived them from the same written source.   Now if the symbols QMt. and QLk. were meant to stand for two related but divergent versions of an oral tradition still in a fluid state, they would be illuminating. But they do not mean this; they stand for documents.  The fallacy is obvious.  Q itself is held to be a written document, because the verbal resemblances between the majority of the parallels between Matthew and Luke are so close as to demand for their explanation the fixity of writing in the common source. But if there are certain passages found in both Matthew and Luke (the Beatitudes or the Lord's Prayer, for example) where the verbal differences between the two versions are so great that they cannot reasonably be supposed to be copied from the same written source, then the only legitimate inference is that one, or possibly both, of these items were not derived from the same written source   to which we have referred the closer parallels.  But, if they were not derived from the same document, they must either have come from oral tradition or from a different document.  A saying which occurs only in Matthew may possibly have stood in Q and have been omitted by Luke; but where a saying occurs in both Matthew and Luke, but Matthew's version is so different from Luke's that the difference cannot be explained as merely editorial, we have clear proof that at least one of the two versions did not stand in the common source.

The fact we have to explain is the occurrence in Matthew and Luke of two sets of parallelism, one set in which the verbal resemblances are so close as to favour, if not actually compel, the conclusion that they were derived from a common written source, and another set in which the divergences are so great that they cannot be explained in that way. And this distinction is not affected by the existence of border-line cases which would be susceptible of either explanation. This two-sided fact is precisely what constitutes the problem we have to solve; and the symbols QMt. and QLk. are discovered on analysis to be merely a means of covering up the phenomena to be explained. Scholars like Mr. Patton have done very valuable service in proving that a number of the parallels cannot be referred to Q. But Q is Q, a document that can be clearly conceived. QMt. is "Q with a difference"; and it may turn out on examination that the difference is just the thing that matters. It may be hard to decide in certain cases whether editorial modification of a saying in Q will or will not account for the differences between the form in which it occurs in Matthew and Luke; but in the last resort we must choose between Q and not-Q. We cannot fall back on QMt. and QLk. as a kind of Limbo for innocent sayings unfortunately disqualified from entering Q.

parallel versions

Wherever the sayings and doings of a remarkable person are preserved in the memory of his followers, different versions of what is substantially the same matter soon become current. If at a later date different individuals in different places conceive the idea of setting down in writing the most interesting or important of the incidents and sayings, which they either remember or can collect from others, four things will inevitably follow.

  1. With each writer the total range of incidents and sayings available will be different; but so also will be the principle in accordance with which each selects from the available material what seems to him of special interest.
  2. Each selection will, therefore, be to a large extent a different one; on the other hand, it would be nothing short of a miracle if the difference were so great that in no case did the same incident or discourse occur in more than one selection.
  3. Where the same item occurs in more than one selection, sometimes it will occur in both in substantially the same form; sometimes two versions will develop which, in the vagaries of oral tradition, will become considerably differentiated.
  4. The sayings having the most universal appeal will be likely to appear in more than one source, and also in the most divergent versions.

This last phenomenon is a notable feature of the Gospels. A comparatively unimportant discourse, like John the Baptist's denunciation of the Pharisees, occurs word for word the same in both Matthew and in Luke, and presumably, therefore, stood in Q, but probably in no other source. But things of outstanding interest like the Beatitudes, the Lord's Prayer, the Lost Sheep, the teaching about Loving Enemies, Forgiveness, or the Strait Gate, are given in strikingly different forms. These then, we infer, stood in at least two sources, and circulated widely in more than one version.

From this it follows that we start out on our investigation with the a priori assumption that we are likely to find numerous cases where the same or similar material stood in more than one source. The assumption is one that justifies itself at once. We cannot move a step without running up against evidence for a considerable amount of overlapping of sources. Indeed, blindness to the evidence for the phenomenon of overlapping sources a blindness artificially induced by the "unconscious assumptions" implied in the "Two Document " nomenclature has, more than anything else, retarded a satisfactory solution to the Synoptic Problem.

The existence of parallel traditions is conspicuous the moment we study the evidence of the non-canonical sources of parables or sayings of our Lord.

  1. I print below the passage from the Epistle of Clement (i. 1 f.), to which I have already referred, with the nearest parallels in Matthew's Sermon on the Mount and Luke's Sermon on the Plain. In view of the express formula of quotation with which Clement introduces the words, it is difficult to believe that it is merely a free rendering of the general substance of mingled reminiscences of Matthew and Luke combined. But, if they are a quotation, they are evidence of the existence in the Church of Rome of a discourse document to some extent parallel to the Great Sermon in Matthew and Luke.
1 Clem.i.1 f. Matt.v.7 etc. Luke vi.36-38

Μάλιστα μεμημςένοι τῶν λόγων τοῦ κερίου Ἰησοῦ οῦς ἐλάλησεν διδάσκων ἐπιείκειαν καὶ μακροθυμίαν. οὕτως γὰρ εῖπεν. Ἐλεᾶτε ἵνα ἐλεηθῆτε, ἀφίετε ἵνα ἀφεθῇ ὑμῖν ὡς ποιεῖτε οὕτω ποιηθήσεται ὑμῖν‧ ὡς δίδοτε, οὕτως δοθήσεται ὕμιν‧ ὡς κρίνετε οὕτως κριθήσεσθε‧ ὡς χρηστεύεσθε, οὕτως χρηστεύσεται ὕμιν‧ ῷ μέτρῳ μετρεῖτε, ἐν αὐτῷ μετρηθήσεται ὑμῶν.

v.7. μακάριοι οἱ ἐλεήμονεςὅτι αὐτοὶ ἐλεηθήσονται.
ἐὰν γὰρ ἀφῆτε τοῖς ἀνθρώποις τὰ παραπτώματα αὐτῶν, ἀφήσει καὶ ὑμῖν ὁ πατὴρ ὑμῶν ὁ οὐράνιος.
μὴ κρίνετε, ἱνα μὴ κριθῆτε
ἐν ᾧ γὰρ κρίματι κρίνετε κριθήσεσθε, καὶ ἐν ᾧ μέτρῳ μετρεῖτε μετρηθήσεται ὑμῖν.
πάντα οὖν ὁσα ἐὰν θέλητε ἱνα ποιῶσιν ὑμῖν οἱ ἀνθρωποι, οὑτως καὶ ὑμεῖς ποιεῖτε αὐτοῖς: οὗτος γάρ ἐστιν ὁ νόμος καὶ οἱ προφῆται.

36. γίνεσθε οἰκτίρμονες καθὼς [καὶ] ὁ πατὴρ ὑμῶν οἰκτίρμων ἐστίν.
καὶ μὴ κρίνετε, καὶ οὐ μὴ κριθῆτε
καὶ μὴ καταδικάζετε, καὶ οὐ μὴ καταδικασθῆτε. ἀπολύετε, καὶ ἀπολυθήσεσθε
δίδοτε, καὶ δοθήσεται ὑμῖν μέτρον καλὸν πεπιεσμένον σεσαλευμένον ὑπερεκχυννόμενον δώσουσιν εἰς τὸν κόλπον ὑμῶν: ᾧ γὰρ μέτρῳ μετρεῖτε ἀντιμετρηθήσεται ὑμῖν.
καὶ καθὼς θέλετε ἱνα ποιῶσιν ὑμῖν οἱ ἀνθρωποι, ποιεῖτε αὐτοῖς ὁμοίως.

  1. The relation of the Gospel according to the Hebrews to that of Matthew is a question that has been much disputed. But we know that it contained a version of the parable of the Talents (with details not unlike parts of the Prodigal Son) and of the injunction "Forgive seven times " (cf. p. 282). It also gave variant versions of the Healing of the Withered Hand and of the story of the Rich Young Man. The resemblances and differences between these and the Synoptic versions can only be explained by the theory of overlapping between the sources of this Gospel and those of the Synoptics. Even if we accept the theory that the document quoted by Jerome was a translation into Aramaic of the Greek Matthew, we must still assume that the text has been influenced by interpolation of parallel versions of these particular sayings current in oral tradition. [The relevant passages from the Gospel according to the Hebrews, from 2 Clement, and, with one exception, from the Oxyrhynchus Logia, are printed at length in the article "Agrapha" in the supplementary volume of Hastings' Dictionary of the Bible.]
  2. The same kind of evidence is afforded by two very ancient interpolations in the text of Matthew. In the Western text, after Mt.xx.28, is inserted a parallel version of the saying about taking the best seats at banquets which is recorded in Lk.xiv.8 f. An exactly similar "Western" insertion is the saying "Signs of the times," Mt.xvi.2-3. This, though found in most later MSS. and included in the Textus Receptus, stands, so far as the early authorities for the text are concerned, on practically the same footing. Both passages are found in D Old Lat.; while the support for "Signs of the times" by C L and the Eusebian canons is of no more weight than that of Φ Syr. (C and Hclmg.) for the addition to Mt.xx.28. Both seem to have been lacking in the oldest Alexandrian, Caesarean, and Antiochene text, being absent from fam. Θ, Syr. S., and Origen's Commentary on Matthew, as well as from B א. But the point I wish to make is that these passages are not harmonistic insertions derived from the text of Luke. For if a later scribe, who had Luke before him, had desired to insert equivalent sayings in Matthew, he would have adhered far more closely to Luke's version. The passages are printed below.
1 Matt.xvi.2-3. Luke .54-57.
Om. B א Syr. S. and C. Arm. Orig.Mt.  

ὀψίας γενομένης λέγετε, εὐδία, πυρράζει γὰρ ὁ οὐρανός‧
καὶ πρωΐ, σήμερον χειμών, πυρράζει γὰρ στυγνάζων ὁ οὐρανός. τὸ μὲν πρόσωπον τοῦ οὐρανοῦ γινώσκετε διακρίνειν, τὰ δὲ σημεῖα τῶν καιρῶν οὐ δύνασθε.

ὁταν ἰδητε [τὴν] νεφέλην ἀνατέλλουσαν ἐπὶ δυσμῶν, εὐθέως λέγετε ὁτι ὀμβρος ἐρχεται, καὶ γίνεται οὑτως‧ καὶ ὅταν νότον πνέοντα, λέγετε ὅτι Καύδων ἔσται, καὶ γίνεται. ὑποκριταί, τὸ πρόσωπον τῆς γῆς καὶ τοῦ οὐρανοῦ οἰδατε δοκιμάζειν, τὸν καιρὸν δὲ τοῦτον πῶς οὐκ οἰδατε δοκιμάζειν‧τί δὲ καὶ ἀφ' ἑαυτῶν οὐ κρίνετε τὸ δίκαιον;

Matt.xx.28. Luke xiv.8-10.
Add. D Φ Old Lat. Syr. C. Hclmg  

ὑμεῖς δὲ ζητεῖτε ἐκ μικροῦ αὐξῆται καὶ ἐκ μείζονος ἔλαττον εἶωαι. εἰσεπχόμενοι δὲ καὶ παρακληθέντις δειπνῆσαι μὴ ἀνακλινεσθε εἰς τοὺς ἐξέχοντας τόπους, μήποτε ἐνδοξότερός σου ἐπέλθῃ καὶ προσελθὼν ὁ δειπνοκλήτωρ εἴπῃ σοι, Ἔτι κάτω χώρει, καὶ καταισχυνθήσῃ. ἐὰν δὲ ἀναπέσης εἰς τὸν ἥττονα τόπον καὶ ἐπέλθῃ σου ἥττων, ἐρεῖ σοι ὁ δειπνοκλήτωρ, Σύναγε ἔτι ἄνω, καὶ ἔσται σοι τοῦτο χρήστιμον.

ὁταν κληθῇς ὑπό τινος εἰς γάμους, μὴ κατακλιθῇς εἰς τὴν πρωτοκλισίαν, μήποτε ἐντιμότερός σου ᾖ κεκλημένος ὑπ' αὐτοῦ, καὶ ἐλθὼν ὁ σὲ καὶ αὐτὸν καλέσας ἐρεῖ σοι, δὸς τούτῳ τόπον, καὶ τότε ἀρξῃ μετὰ αἰσχύνης τὸν ἐσχατον τόπον κατέχειν. ὁ κεκληκώς σε ἐρεῖ σοι, φίλε, προσανάβηθι ἀνώτερον: τότε ἐσται σοι δόξα ἐνώπιον πάντων τῶν συνα νακειμένων σοι.

One has only to read them through side by side to see that the verbal agreements between the two versions are almost nil, and can only be accounted for on the hypothesis that the interpolations are drawn from a tradition independent of Luke. Probably they are excerpts from the primitive discourse document of the local Church in which the interpolator worked. The MS. evidence would favour the view that both readings originated in Rome. In that case they may well be fragments of the same document or catechetical tradition from which Clement quotes.

  1. The Oxyrhynchus Logia contain some sayings of our Lord, which have a close resemblance to sayings found in the Gospels, and others, which exhibit a remarkable combination of resemblance and divergence. The definite citations of words of Christ in the second century homily known as 2 Clement exhibit the same phenomena of versions of sayings more or less parallel to those contained in the Canonical Gospels. Some scholars think the sayings preserved in the Oxyrhynchus Logia and by 2 Clement are all from the same lost Gospel "according to the Egyptians"or "according to the Hebrews." If so, they prove that the sources of that Gospel to a considerable extent overlapped with those of the Synoptics. If, on the other hand, the sayings in question come from more than one lost document, the evidence for parallel traditions is further multiplied.

the overlapping of sources

Let us now, from the standpoint of this evidence, re-examine certain phenomena in the Canonical Gospels.

  1. It has long been recognised that Q and Mark to some extent overlap. The question whether in these passages Mark is directly or indirectly dependent on Q has been discussed above (p. 187 ft.), and the conclusion was reached that it is more probable that Mark and Q represent two independent traditions.
  2. We have also seen that there is similar overlapping between Mark and the narrative material in the source L, since Luke evidently had before him versions of the Rejection at Nazareth, the Call of Peter, the Anointing, and of the whole story of the Passion, which on the whole he preferred to the accounts which he found in Mark.
  3. A third clear case of overlapping is seen in the parables of Matthew and Luke. It is sometimes doubtful whether we ought to call a particular saying a short parable or an extended illustration, but taking the list on p. 332 plus those named here we count fifteen parables in Matthew and twenty-three in Luke. Two of these, the Sower and the Wicked Husbandmen, are derived from Mark [The Mustard Seed, of course, stands in Mark as well as Q.]; two more, the Mustard Seed and Leaven, certainly stood in Q. There remain to be accounted for two collections of parables though, of course, in speaking of them as "collections" I do not wish to imply that they were derived from different sources from the rest of the peculiar matter in the Gospels where they occur. These collections number respectively eleven and nineteen. But they overlap to the extent of three parables, since each collection includes a version of the Lost Sheep, the Marriage Feast ( = the Great Supper), and the Talents ( = Pounds). [Matthew's Marriage Feast (the King's Son) (x.1 ff.) is really two parables. Verse 2, or words to that effect, has evidently been omitted before verse 11. Repeat verse 2 here, and verses 11-14 are seen to form the second half of one of those pairs of "twin parables" enforcing a different aspect of the same general moral, so characteristic of our Lord's teaching. Without such emendation the second half is pointless. How could the man, just swept in from the highways, be expected to have on a wedding garment?] But though these three parables occur in both Matthew and Luke, they do so in such very different forms that the supposition that they were derived from Q postulates too large an amount of editorial manipulation of that source.

But the ingenuity that attempts to derive them from a single written source is wholly misdirected. Given that two different persons set about collecting parables of our Lord, and that one of them succeeded in finding eleven and the other as many as nineteen. Would it not be an astounding circumstance if the two collectors never happened to light upon the same parable? The remarkable thing is, not that the two collections have three parables in common, but that they have only three.

Thirteen years ago I myself, under the malign influence of the "unconscious assumptions" of the Two Document Hypothesis, argued that these three parables occurred in Q. [Oxford Studies, p. 197 ff.] But one day, while I was meditating on the curious fact that the moral which Matthew draws from the parable of the Lost Sheep is quite different from that drawn by Luke, it occurred to me that this is precisely what one would expect if the two versions had been handed down in two different traditions. People so often remember a story or an illustration, but forget the point it was told to illustrate. Then I turned to Harnack's famous reconstruction of Q. [ A. Harnack, The Sayings of Jesus, p. 93 (Williams and Norgate, 1908).]  I found that, in order to derive both versions from Q, he had to maintain that the saying

There shall be joy in heaven over one sinner that repenteth
more than over ninety and nine just persons which need no repentance

was an editorial addition.

The scales fell from my eyes.

No saying attributed to Jesus can have struck those who first heard it as so utterly daring as this. I reflected that, if a man of Harnack's insight can be driven by the logic of his premises to the conclusion that such a saying is an editorial addition, there must be something wrong about the premises. Then it dawned on me that the assumption on which he—and I too—had been working was fundamentally false. Even if the differences between the versions did not demonstrate, antecedent probability would lead us to expect, that two different collections of parables would certainly overlap.

I proceeded at once to re-examine the parable in a Synopsis, and I saw at once that if, instead of mechanically counting the number of Greek words common to the two versions, one asked which of the really significant words were found in both versions and which in only one, the conclusion that the versions were independent was confirmed. The words which are found in both versions are the words without which the story could not be told at all "man," "sheep," "go," "find," "rejoice,"the "I say unto you" (which is the regular formula for pointing the moral in our Lord's teaching), and the three numerals, 100, 99, and 1, which since 100-1 = 99 would be inevitable in any version. But where the versions can differ, they do so.
For Matthew's "if it happen to a man," Luke has "what man of you?";
for "be gone astray" (passive), "having lost " (active);
for "into the mountains," "in the wilderness";
for "seek," "go after" ;
for "layeth it on his shoulder rejoicing," "rejoiceth over it."
Luke adds the calling together of friends and neighbours, about which, Matthew is silent, and the saying about the joy in heaven over the sinner repentant;
while Matthew, instead of this, points the moral,

Even so it is not the will of your Father
that one of these little ones should perish.

The differences between the two versions of the parable of the Lost Sheep are as nothing compared to the differences between the other two pairs, the Marriage Feast = the Great Supper, and the Talents = Pounds. But since Matthew has eleven and Luke nineteen parables, and twenty-seven of these thirty must have been derived from two quite different cycles of tradition, the probability that the two cycles overlapped to the extent of including divergent versions of at least three parables is a high one.

It appears, then, that the occurrence in overlapping sources of parallel versions of the same saying is characteristic alike of canonical and extra-canonical reports of the teaching of Christ. All analogies, therefore, are in favour of the hypothesis that Q and M also would at some points overlap.

Matthew's method of conflation

When an editor combines sources that cover the same ground along some part of their extent, he has a choice of two methods. He can either accept the version given by one source and ignore the other, or he can make a careful mosaic by "conflating" the two. We noticed in a previous chapter (p. 187 ff.) that, when the same saying occurs in both Mark and Q, Luke commonly accepts the Q version and ignores Mark's; Matthew, on the other hand, usually conflates Mark and Q, though with a tendency to abbreviate. Now, if it was Luke himself who first combined Q and L, and if on this occasion he followed his later practice of choosing one of two versions rather than combining them, all traces of any overlapping there may have been between Q and L will have been eliminated. On the other hand, if Matthew pursued as between M and Q the same method of conflation which he used where Q and Mark overlap, some traces of the double version will still remain. The detection of these is the immediate goal of our inquiry.

In order, however, to do this we must first study the way in which Matthew conflates. It will appear that he not only pieces together the substance of sayings that occur in two different sources, but he combines minute points of difference in their expression of the same thought.

The way in which the wording of Matthew's parable of the Mustard Seed conflates the versions of Mark and Luke is particularly instructive. In the parallels printed below, words found in both Matthew and Mark, but not in Luke, are printed in heavy type; words found in Matthew and Luke, but not in Mark, are underlined. Words found in all three are in red.

Mark iv. 30-32. Matthew.i.31-32. Luke i.18-19.

30. καὶ ἐλεγεν, πῶς ὁμοιώσωμεν τὴν βασιλείαν τοῦ θεοῦ, ἢ ἐν τίνι αὐτὴν παραβολῇ θῶμεν;
ὡς κόκκῳ σινάπεως, ὃς ὁταν σπαρῇ ἐπὶ τῆς γῆς, μικρότερον ὂν πάντων τῶν σπερμάτων τῶν ἐπὶ τῆς γῆς,
32. καὶ
ὁταν σπαρῇ, ἀναβαίνει καὶ γίνεται μεῖζον πάντων τῶν λαχάνων καὶ ποιεῖ κλάδους μεγάλους, ὡστε δύνασθαι ὑπὸ τὴν σκιὰν αὐτοῦ τὰ πετεινὰ τοῦ οὐρανοῦ κατασκηνοῦν.

31. ἀλλην παραβολὴν παρέθηκεν αὐτοῖς λέγων, ὁμοία ἐστὶν ἡ βασιλεία τῶν οὐρανῶν κόκκῳ σινάπεως, ὃν λαβὼν ἀνθρωποςσπειρεν ἐν τῷ ἀγρῷ αὐτοῦ:
32. ὃ
μικρότερον μέν ἐστιν πάντων τῶν σπερμάτων, ὁταν δὲ αὐξηθῇ μεῖζον τῶν λαχάνων ἐστὶν καὶ γίνεται δένδρον, ὡστε ἐλθεῖν τὰ πετεινὰ τοῦ οὐρανοῦ καὶ κατασκηνοῦν ἐν τοῖς κλα δοις αὐτοῦ.

18. ἐλεγεν οὖν, τίνι ὁμοία ἐστὶνβασιλεία τοῦ θεοῦ, καὶ τίνι ὁμοιώσω αὐτήν;
ὁμοία ἐστὶν κόκκῳ σινάπεως, ὃν λαβὼν ἀνθρωπος ἐβαλεν εἰς κῆπον ἑαυτοῦ, καὶ ηὐξησεν καὶ ἐγένετο εἰς δένδρον, καὶ τὰ πετεινὰ τοῦ οὐρανοῦ κατεσκήνωσεν ἐν τοῖς κλάδοις αὐ τοῦ.

In the above parallels we must ignore as irrelevant for such comparisons all occurrences of the word "and," the verb "to be," the definite article, and all pronouns; there remain 31 words in Matthew's version. Of these only 7 are his own; 7 occur also in Mark and Luke; 10 in Mark; 7 occur in Luke but not in Mark, and so presumably stood in Q. Now if Mark had been lost, every one would have explained the verbal differences between Matthew and Luke as due either to editorial amplification by Matthew or to abbreviation by Luke. As it is, we see that practically every word in Matthew is drawn from one or other of his two sources. But the differences between the Marcan and the Lucan (i.e. the Q) version of the parable are entirely unimportant. They in no way affect the general sense, and no one antecedently would have expected that Matthew would take the trouble to combine the two versions. The fact that he has done so where so little was to be gained is thus a very important revelation of the care he would be likely to take when combining sources containing differences of real interest.

Another example, hardly less illuminating, of the almost meticulous care with which Matthew conflates Mark and Q is the discourse Mt.x.9-15.   This example is somewhat complicated by the existence of a fourth version, Lk.ix.3-5.  This is mainly from Mark, but its differences from Mark seem to arise from conflation with the same Q discourse as that best preserved in Lk.x.4 ff.; Lk.ix.3-5, however, seems to retain a word or two from Q which has been modified in Lk.x.4 ff.  This complication, and the fact that Matthew repeats the words "entering," "house," "peace," and "worthy" more than once, would make a merely statistical statement misleading.  The notable thing is that the only real additions he has to make are the words "gold" (verse 9) and "Gomorrha" (verse 15), both of which are due to verbal association "Gomorrha" is suggested by Sodom, and "gold" by silver. Gold is hardly original; it was not a commodity which those for whom the words were first intended needed exhortation not to carry, though copper or silver might be.  Apart from these two, obviously editorial, additions, there is not a word in any way significant which is not to be found either in the Marcan or one of the Lucan parallels.

There is another point to notice.
The necessity of conflating the Marcan and the Q versions has led Matthew entirely to rearrange the order of the sentences in Q, which we may presume to be preserved in Lk.x.4 ff. There are only two other possible Q passages where such a redistribution of sentences within a single section occurs (I do not speak of diversity in the order of complete sections dealing with separate topics, which is quite another matter). These are Mt.v.38-48 =, "Love your enemies" and the Denunciation of the Pharisees, Mt.xi.1-36 = Lk.xi.39-52. We shall see later (p. 252 f.) that here also Matthew may have rearranged the order of Q, preserved by Luke, in order to conflate with a parallel version from another source.

A study of these and other cases, similar though not quite so striking, shows that, wherever we have reason to suspect that Q and M overlap, we must insist on the probability that the divergence between the two versions was originally greater than that between the parallels as they now stand in Matthew and Luke. Indeed, Matthew's habit of conflating the actual language of parallel sources compels us to formulate a new principle of Synoptic criticism. Wherever parallel passages of Matthew and Luke exhibit marked divergence, editorial modification of Q is a less probable explanation than conflation of Q by Matthew with the language of a parallel version. I need not pause to point out the havoc wrought by the formulation of this principle in various critical reconstructions of Q my own included—which are based on exactly the opposite assumption. [For similar "conflation" in Mediaeval documents see C. Plummer, Expositor, July 1889.]

the sermon on the mount

In view of the evidence that the overlapping of sources is a vera causa, and of the principle deduced above from a study of Matthew's method of conflation, let us explore the hypothesis that there is overlapping between Q and M. This, unless I am altogether mistaken, will lead to results of a highly illuminating character. In particular it will explain those well-known difficulties concerning the composition of the Sermon on the Mount to which the Two Document Hypothesis has never been able to give a really satisfactory answer.

The Sermon on the Mount (Mt.v-vii.) is four times as long as Luke's Sermon on the Plain (; but there are two considerable sections of it which, though absent from the Sermon on the Plain, occur in Luke scattered in different contexts. These show such close verbal parallelism to Matthew that they must certainly be referred to Q ( = Lk.xi.34-36, xvi.13, .22-31 and Mt.vii.7-11 = Lk.xi.9-13). These create no difficulty; Matthew has obviously inserted them in their present context in accordance with his practice of AGGLOMERATING, i.e. of collecting into large discourses all the available material dealing with the same or related topics. [Curiously enough, apart from the final Beatitude, and the House on the Sand, all the material in the Sermon on the Mount, which is certainly from Q, comes in the block vi.22-vii.12; and all but two verses in this block are Q.]

Both sermons begin with Beatitudes, both end with the parable of the House built upon the Sand, and to all but six of the intervening verses in Luke's Sermon there are parallels, more or less exact, in the Sermon on the Mount; and both are followed by the story of the Centurion's Servant. The natural inference is that Q contained a Great Sermon followed by the story of the Centurion's Servant. But on closer study it appears that the Sermons in Matthew and Luke can be derived from a single written source only if we postulate an almost incredible amount of editorial freedom in rewriting portions of the original. Thus Matthew has nine (originally, perhaps, eight)  Beatitudes, all but one in the third person; [Verses 4 ("they that mourn") and 5 ("the meek") are transposed by 33 D a c k fam. Θ Syr. C. Orig.Mt.. Transposition results when the next copyist inserts a sentence written in the margin in the wrong place. But, though a passage thus inserted may replace an accidental omission, it may be an interpolation. I incline to agree with Harnack that Mt.v.5 is an interpolation from Ps.xxxvii.11, against Dr. Charles, who, in his The Decalogue (T. & T. Clark, 1923), argues that verse 4 is the interpolated verse, through assimilation from Luke.] Luke has four Beatitudes balanced by four corresponding Woes, all in the second person. Only the final Beatitude in Mt.v.11-12 ( =, unlike the rest, which are in the third person, is in the second person like all the Beatitudes of Luke, and is almost verbally identical with the Lucan parallel. This divergence would not be un­natural if they followed two independent oral traditions of the same discourse; it is not plausibly explained as the result of editorial modification of a written source, for we can check the editorial methods of the authors by studying their handling of the sayings of our Lord which they derive from Mark. Besides the Beatitudes there are in the two Sermons other parallels where the degree of verbal resemblance is really not much greater, for example the two very different versions of the saying "Lord, Lord," Mt.vii.21= Again, in the striking section "Resist not evil. ... Love your enemies " (Mt.v.38-48 =, not only are there considerable diversities of language, but the order in which the component sayings are arranged is entirely different, which, as the example of Mt.x.9 ff. showed, suggest conflation of two sources. Indeed, there are only two considerable passages which occur in both sermons, i.e. "Judge not" (Mt.vii.1-5 =, 41-42) and the House on the Sand (, which can, without postulating a good deal of editorial modification, be explained as being entirely derived from a single common written source.

Let us now try the simple experiment of deducting from the Sermon on the Mount just these passages, which, on account of their close resemblance to parallels in Luke, can with the maximum of probability be assigned to Q.  What remains more than two-thirds of the whole reads like a continuous and coherent discourse. 
Most of it is peculiar to Matthew; but some passages, for example "Love your enemies" and the Lord's Prayer, have parallels in Luke sometimes within, sometimes outside, the Sermon on the Plain.  But these parallels have no more than that general resemblance which one would expect in divergent traditions of the same original saying. All the phenomena, however, can be satisfactorily explained by the hypothesis that Matthew is conflating two separate discourses, one from Q practically identical with Luke's Sermon on the Plain, the other from M containing a much longer Sermon. Both Sermons opened with four Beatitudes.  The Sermon in Q contained the four Blessings in the second person, as in Luke; that in M gave four in the third person, corresponding to Mt.v.7-10.  The Q Beatitude,

Blessed are ye when men shall ...
persecute you ... for My name's sake (Mt.v.11-12),

is a doublet of that in Mt.v.10,

Blessed are they which are persecuted for righteousness' sake,

which stood in M; otherwise the two sets of four do not overlap. Matthew has simply added the two sets together, changing the person and slightly modifying the wording in three of those he takes from Q.
[I owe several points in this analysis to suggestions by Prof. Dodd.]
Mt.v.5, "the meek," is, as the transposition in the MSS. suggests, an early interpolation from Ps.xxxvii.11. The four Woes in Luke vi.24 ff. may have stood in Q and been omitted by Matthew. His explanatory additions to the Blessings, on the Poor (+ in spirit), and on those that hunger (+ after righteousness), show that he might well have thought the denunciations of the "rich," and the "full" (, open to misunderstanding; poverty and hunger as such have no ethical value.

The Sermons in Q and M occasionally overlapped, e.g. in the section on Loving Enemies [Rom..14, "Bless them that persecute you," etc., suggests that various summaries of this part of our Lord's teaching were current.], Mt.v.38-48; the variation in order between the parallels in Matthew and Luke is here very marked, and wherever this happens (cf. p. 248), judging from the way in which the editor of Matthew deals with overlapping of Mark and Q, we suspect that there has been a certain amount of conflation. Hence the Q and M versions of any saying that occurred in both Sermons would in the original sources have shown greater divergence than do the present texts of Matthew and Luke. Having thus conflated the two Sermons from Q and M, Matthew proceeded to add to them certain other passages of Q, which Luke gives later in his Gospel in what is more likely to be their original context in that source. The hypothesis of a summary of Christian teaching intended for catechetical instruction, current in oral tradition in more than one form, has often been invoked to account for the combined phenomena of resemblance and difference between the versions of the Great Sermon in Matthew and in Luke. But as usually presented it goes shipwreck on the fact that, in the source used by both Matthew and Luke, the story of the Centurion's Servant follows immediately after the Great Sermon. That difficulty disappears if, instead of supposing that Matthew and Luke had each a different version of the same Sermon, we suppose that Matthew had before him two documents, Q which contained both the Sermon on the Plain and the Centurion's Servant, and M which gave a substantially different version of the Sermon, but did not include the Centurion's Servant. The idea of conflating the two would be inevitably suggested to Matthew by the fact that both Sermons began with Beatitudes and also that they overlapped at certain other points.

We proceed to consider the long discourse Mt.xi., the Woes to the Pharisees. This is, next to the Sermon on the Mount, the longest connected discourse of which both the Matthean and the Lucan versions (Mt.xi.1-36 = Lk.xi.37-52) cannot be referred to a single written source without raising great difficulties.
Matthew's is much the longer version, and it reads like an early Jewish Christian polemical pamphlet against their oppressors the Pharisees. No doubt it is largely based upon a tradition of genuine sayings of Christ, but we cannot but suspect that it considerably accentuates the manner, if not also the matter, of His criticism of them. Indeed it is the one discourse of our Lord, which, from its complete ignoring of the better elements in a movement like Pharisaism, it is not easy to defend from the accusation made by students of Jewish religion of being un­sympathetic and unfair. Now it is quite commonly assumed as almost self-evident that Matthew's version stood in Q and that Luke's is an abbreviated reproduction of the same source. But there are three considerations that give us pause.

  1. The divergence between the parallels is well above the average in wording and it is accompanied by a great variety in the order—a signpost for conflation (p. 248).
  2. There is a fundamental differ­ence in structure between the two discourses.
    The core of the discourse in Matthew is the seven times repeated

    Woe unto you, Scribes and Pharisees.

But in Luke what we have is three Woes against Pharisees followed by three against Lawyers.

  1. It is to be noted that quite the most striking of the very few cases in the Gospels where the diversity between Matthew and Luke can be plausibly accounted for by independent translation from Aramaic occur in this discourse.
    [Wellhausen suggests that דכ' (cleanse) misread as זכ' (give alms) would account for Luke's

    τά ἔνοντα δότε ἐλεημοσύνην

as compared with Matthew's

καθάρισον πρῶτον τὸ ἐντός.]

The fact that Luke's version of the discourse, xi.37-52, comes in the middle of a section of which the rest is certainly derived from Q, makes it probable that his version stood in that document and that Matthew has again conflated a discourse of Q with one on the same topic which came to him in M. But here, again, the very fact that Matthew's version is a conflation of Q and M means that Mt.xi. as it now stands bears a much closer resemblance to Lk.xi.37-52 than did the original discourse that stood in M. Yet again, Matthew, besides placing the discourse in a Marcan context, adds to it a few words from Mark, e.g. πρωτοκλισίας κτλ., Mt.xi.6 = Mk..39. Finally, we must notice that Matthew has completed his structure by appending xi.37-39, the Q saying, "Jerusalem, Jerusalem," which occurs in what to me looks a far more original context in Lk.i.34-35.

Judaistic tendency of M

Mt.x.9-16 is, as we have already seen, a most careful com­bination of the Charge to the Twelve ( and the Charge to the Seventy (Lk.x.3-12) which we assign to Q. Besides this, the chapter contains other sections derived from either Mark or Q. Mt.x.17-22 seems to have been transferred by the editor from Mk.i.9-13.
Mt.x.36-38 seems to be from Q, while x.40, 42 are from Mark. Only about half a dozen verses remain which are without close parallels in either Mark or Luke. We ask whence were these derived. Much the most striking are the words that precede the conflated discourse:

"Go not into any way of the Gentiles,
and enter not into any city of the Samaritans: 
but go rather to the lost sheep of the house of Israel."

There is, I think, a close connec­tion of thought between this opening and the words that conclude the first half of the discourse,

"Ye shall not have gone through the cities of Israel,
till the Son of man be come."

This verse appears to be intended to give a reason for the previous prohibition to preach to Gentiles or Samaritans. It is not that Gentiles cannot or ought not to be saved, but the time will not be long enough to preach to all, and Israel has the first right to hear.  But if I am correct in this interpretation, the two passages must originally have stood much closer together.  They look like the beginning and end of a Judaistic version of the Charge to the Twelve, the wording of which has taken the precise form it now bears under the influence of the controversy about the Gentile Mission which almost split the early Church. [Schweitzer argued, from Mt.x.23, that Christ expected the Parousia before the return of the Twelve from their preaching tour; but the words clearly reflect a situation that did not come into existence till the Missionary Journeys of Paul. Incidentally, I may remark that Schweitzer's whole argument depends on the assumption that Mt.x. is word for word an exact report of what was said at the time. The demonstration I have given above, that Mt.x.5-23 is a late conflation of at least two sources, Mark and Q, would alone be a sufficient refutation of his argument.]  The question then arises, Did these words stand in Q and form the original beginning and end of the discourse which Luke gives as that to the Seventy?  Or does Mt.x.5-8, 23, with the possible additions of x.24, 25, 41, represent a short Judaistic charge, which Matthew has conflated with the versions given by Mark and Q?  If we elect for the former alternative, we must say that Luke, convinced that a command of the Lord not to go to Gentile or Samaritan could not be genuine, has left out the words.  If we choose the latter, it follows that Q in its original form was a document emanating from the Judaistic section of the Church.

Against the view that Q was a Judaistic document two considerations may be urged:

  1.   The occurrence in it of the pro-Gentile incident of the Centurion's' Servant, and the saying about the Law and the Prophets being until John, which, whatever its original meaning, certainly lends itself to the view that the Old Law was in a sense superseded by the Gospel. To this it may be objected that since in Matthew these and similar sayings occur side by side with ones of a Judaistic tenor, the same thing may have happened in Q. But to this I would reply that it is not very likely that the author of a primitive document would put side by side sayings implying contrary rulings on what at the time he wrote was a highly controversial issue; it is quite another matter for a later writer, very conservative as Matthew is in his use of his sources, to include contrary sayings found in two different ancient documents, especially as the controversy in question had by that time largely died down.
  2. Judaistic sayings in Matthew only occur in contexts which on other grounds we should refer to M, or where there is evidence of conflation between Q and another source. In all these Judaistic passages it is difficult not to suspect the influence of the desire of the followers of James to find a justification for their disapprobation of the attitude of Paul, by inventing sayings of Christ, or misquoting sayings which, even if authentic, must originally have been spoken in view of entirely different circumstances. The followers of every great leader have always quoted his sayings in the next generation to justify their own attitude in circumstances quite different from his; and where there exists no written or printed record to check their original form it is easy for the actual wording, as well as the application, of the sayings to become changed.

The first of these passages is Mt.v.17-20, which defines the relation of Christianity to the Law. The saying,

"Whosoever therefore shall break one of these least commandments,
and shall teach men so,
shall be called least in the Kingdom of Heaven",

is sharply contrasted with

"Whosoever shall do and teach men so,
he shall be called great in the Kingdom."

This reflects the attitude of the Jewish Christians who, while barely tolerating the proceedings of Paul, regarded as the pattern Christian, James, surnamed the Just, because his righteousness, even according to the Law, did exceed that of the Scribes and Pharisees. [It is possible that the passionate protest,

I am the least of the apostles ...
but I laboured more abundantly than they all,

1 Cor.xv.9 ff.,

has a reference to a description of him and his work by the Judaisers in words not dissimilar to those in the text.] It is to be remarked that this passage does not come in that part of the Sermon on the Mount which we have referred to Q.

The same idea is still more clearly enforced in Mt.xi.2-3,

"The Scribes and Pharisees sit in Moses' seat:
all things therefore whatsoever they bid you,
these do and observe."

Here we have attributed to our Lord an emphatic commandment to obey, not only the Law, but the scribal interpretation of it. That is to say, He is represented as inculcating scrupulous obedience to that very "tradition of the elders" which He specifically denounces in Mk.vii.13. But here again we have already, on other grounds, seen reason to suppose that Matthew's version of this discourse was derived largely from M.

The section (Mt.xviii.15-22) "If thy brother sin against thee ... till seventy times seven" differs in wording from Lk.xvii.3-4 so much that it is not likely that both passages were taken from Q; especially as we know of another version of this particular saying in some ways intermediate between those of Matthew and Luke preserved in the Gospel according to the Hebrews (cf. p. 282). It must therefore be assigned to M. Now an important little point, affording confirmatory evidence that the sayings of a Judaistic type are connected with M rather than with Q, is the fact that on examination it appears that this saying, as it occurred in M, was set in a Judaistic context. Only here, and in the passage "Thou art Peter," does the word "Church" occur in the Gospels; and the word "Church" in this context clearly means the little community of Jewish Christians. In a Gentile community tradition would surely have modified the form of the injunction "If he refuse to hear the Church, let him be unto thee as the Gentile and the publican."

[As early as Hermas, ecclesiastical writers use the term "Gentiles" as equivalent to pagans; but this usage implies a time when the controversy whether Gentiles could be admitted to the Church on equal terms with Jews had long ago been settled, τὰ ἔθνη occurred in Q ( = Lk.vii.30), but not in such, an invidious sense as the ἐθνικός of Mt.xviii.17 and Mt.v.47. Luke, however, tones it down in vii.10 by adding τοῦ κόσμου.]

It might be argued that a similar "tendency" appears in the famous "Thou art Peter" (Mt.xvi.18 ff.). In the Jewish idiom, "I will give unto thee the keys of the Kingdom of Heaven" means "I appoint thee my Grand Vizier"; and "to loose" and "to bind" are technical terms for declaring permissible or the reverse particular lines of conduct in the light of the obligations of the Law. The passage, in the form in which we have it, is an emphatic declaration that Peter is the Apostle who on these points could speak with the authority of Christ. What our Lord really said to Peter, and what at the time of speaking He meant by it, is an entirely different question; and it is not one to which we are likely to find an answer with which everybody will be convinced. But whatever the words meant as originally spoken, it is hard not to suspect that they have since been modified by some controversy between the followers of different leaders in the early Church. But to my mind it is less likely to have been the controversy between the party who said "I am of Peter" and the admirers of Paul, than that between the extreme Judaisers who exalted James to the supreme position and the intermediate party who followed Peter.
[The Clementine Homilies open with a letter from Peter to James beginning,

Peter to James,
the lord and bishop of the holy Church,
under the Father of all, through Jesus Christ.

This is followed by one from Clement,

Clement to James,
the lord (or lord's brother) and the bishop of bishops,
who rules Jerusalem ... and the churches everywhere.

The Homilies probably date ± 180, but in this particular regard must represent a party feeling of an earlier period.]
 In that case "Thou art Peter" will have been derived, not from M, but from the local traditions of Antioch the headquarters of this intermediate party.  But we shall refer to M the doublet of this saying, Mt.xviii.17, which confers the power "to bind and loose" upon the Ecclesia, that is, on the righteous remnant of the People of God, of which the Jerusalem Church was the natural headquarters and shepherd.

overlapping of Mark and M

Seeing there is evidence of the existence of a source evidently emanating from a Judaistic circle, we must not overlook the possibility that there would be overlapping between it and Mark as well as between it and Q. And it is the fact that the occurrence of parallel versions of the same incident in Mark and M would explain three cases where Matthew's account appears to be in some ways more original than Mark's.

  1. Matthew's section on Divorce (Mt.xix.3-12) is both more naturally told and more closely related to Jewish usage than the parallel in Mark (Mk.x.2-12). [Cf. R. H. Charles, The Teaching of the New Testament on Divorce, p,( 85 ff (Williams and Norgate, 1921).] The words "for every cause" in the question put by the Pharisees look more original, since they imply that the point submitted to the reputed Prophet in regard to the grounds of divorce was one actually debated at the time between the schools of Hillel and Shammai. So does our Lord's reply, referring them for an answer to the fundamental principle stated in Genesis, "They two shall be one flesh." The reference to the law of divorce in Deuteronomy comes more appropriately, as in Matthew, in their reply to Him than, as in Mark, as our Lord's original answer. And, finally, Matthew's arrangement makes His final rejoinder, that this was merely permissive, more effective.
  2. In the story Mt..9-13 told, not for the sake of the healing miracle, but to illustrate our Lord's attitude to the Sabbath Matthew adds to Mark the detail "a sheep in a pit."  If we compare with this the addition "ox in a pit" in the similar story in Luke (xiv.1-6), we shall be inclined to attribute it to conflation with another version rather than to editorial expansion.
  3. The account of the Syrophenician woman, as given by Matthew, is made, by an addition of the two and a half verses (Mt.xv.22b-24) (which suggest very great reluctance on the part of our Lord to heal a Gentile), very much more Judaistic than the version given by Mark (vii.24-30). [The reading of Syr. S. is even more Judaistic: "I have not been sent save after the flock, which hath strayed from the House of Israel."]

But Divorce, the Sabbath, and the position of Gentiles were all burning questions, especially among Jewish Christians. Hence we should expect that sayings or stories which could be quoted as defining Christ's attitude towards them would be current at a very early time in nearly every Church and most certainly in the Church of Jerusalem. It seems likely, then, that in these three instances Matthew had before him a parallel version in M. But in each case he tells the story in the context in which it occurs in Mark. Probably, then, he takes Mark's version as his basis, adding only a few notable details from that of M. Thus only fragments of the M version are likely to have been preserved, and its original form may have differed considerably from Mark. Hence here, as so often, we cannot reconstruct the M version.

In view of the evidence submitted in this and the two pre­ceding sections, it is, I think, clear that Matthew made use of a cycle of tradition of a distinctly Judaistic bias which to some extent ran parallel to the cycles preserved in Mark, in Q, and in L. If we suppose that the whole of the Parable and Discourse material peculiar to Matthew, plus the sections commented on above, came from a single source, it would be of much the same length as Q; and the proportion of this source paralleled in other sources would not be greater than the proportion of Q that is paralleled by Mark. For the view that the whole of this material came from a single source the amount of evidence that can be produced is small. All that we can say is that, while only a few passages are Judaistic in the party sense, the whole of it is redolent of the soil of Judaea; that it is the kind of collection we should expect to emanate from Jerusalem; and, lastly, that it is hard to account for the fact that so very little tradition of any value has survived outside the Four Gospels, unless we suppose that the tradition of the Church of Jerusalem, which we should expect to be quite exceptionally rich, is incorporated in one or other of those Gospels. That Matthew made use of a source or sources which were in some respects parallel to Q and L, I regard as proved; that this material, along with, at any rate, the bulk of his peculiar matter, was the cycle of tradition of the Church of Jerusalem, is in no sense proved; but it seems more probable than any alternative suggestion.

the great discourses of Matthew

In past times more than one critic has put forward the hypothesis that five great discourses of Mt.v.-vii., x., i., xviii., xxiv.-xxv. [Some think that chap, xi. should be regarded as part of the same discourse as xxiv.-xxv., the saying,
"Thy house is left unto thee desolate" (xi.38), being interpreted of the Temple, whose destruction is the theme of chap.xxiv.]
were taken over by him practically unaltered from an earlier source. One great objection to this theory is that the four lesser ones seem largely built up of material derived from Mark. But, of course, if there is reason to believe that M contained material closely parallel to parts of Mark, that objection is shaken. Accordingly I have felt it incumbent on me to reinvestigate this hypothesis. The result of such reinvestigation is distinctly unfavourable to its acceptance. But as the conclusion come to on this point has an important bearing on any reconstruction of Q we may attempt, I will briefly lay the facts before the reader.

The chief attraction of this hypothesis is that it would explain the occurrence five times, after each of these great discourses, of the formula,

It came to pass when Jesus had finished these sayings that. ...

It has been pointed out that the number five is a standard number in literary usage, both Jewish and early Christian. There are five books of the Law and of the Psalms, five original divisions of the Rabbinical works the Megilloth and the Pirqe Aboth, and Papias wrote "Interpretations of the Sayings of the Lord," divided into five books. It has been suggested [Hawkins' Hor. Syn.2 p. 163 ff.] that the above formula is the remains of a colophon, comparable to "The prayers of David the Son of Jesse are ended" which appears to have once marked the conclusion of an earlier collection included in the Psalms (cf. Ps.Ix.20). To this I would reply that a colophon, though appropriate at the end of a volume, would seem a trifle ridiculous at the end of collections of sayings not longer than Mt.x., i., or xviii. Again, the formula has really no resemblance to a colophon; its emphasis is not on the "Here endeth" but on "Here beginneth"; it is a formula of transition from discourse to narrative. Nor does its occurrence five times in Matthew constitute evidence that it occurred just that number of times in his source; for "repetition of formulae" is one of the notable characteristics of his Gospel. [Cf. ibid. p. 168.]
It is just possible that Matthew may have found the formula in
Q, for a phrase rather like it occurs after Luke's Sermon on the Plain in a context parallel to the occurrence of the formula in Matthew after the Sermon on the Mount (Lk.vii.1 = Mt.vii.28). But, if it stood in Luke's copy of Q, there also it would have done so as a formula of transition from discourse to narrative; for in Luke it occurs between the Great Sermon and the story of the Centurion's Servant. It would seem likely, then, that Matthew found the formula in Q, and thought it a convenient one to repeat whenever he had occasion to mark a similar transition from a long discourse to narrative.

But, whatever may be the origin of this formula, there are insuperable difficulties in the way of supposing that these five great discourses stood in M in anything like their present form.

  1. The Sermon on the Mount we have already discussed.
    We have seen that perhaps two-thirds of it did stand in M as a continuous discourse, but that it was by no means the equivalent of Mt.v-vii., for Matthew has inserted into it large sections of Q. [There are a few short sayings in it which have parallels in Mark, but whether these have been inserted by Matthew from Mark, or whether in these instances there were sayings in Q or M similar to Mark it is not possible to determine.] An examination of the other discourses yields even clearer results.
  2. The discourse part of Mt.x. opens, as we have seen above (p. 254 f.), by conflating a Mission Charge from the three sources, Mark, Q, and M (Mt.x.5-16, 23). Verses 24-25 and 41, being peculiar to Matthew, are probably M. Everything in 36-39 has parallels in Luke in different contexts, but as these are not all equally close, we may leave it an open question how much of this section is from Q and how much from M. But there are three other passages that have close parallels in Mark (Mt.x.17-20, 40, 42). In the first and third we may be pretty certain that Matthew is dependent on Mark. Thus the saying

    Stand before governors ... as a testimony unto them.
    And the gospel must first be preached to the Gentiles,

in its Marcan form and context (Mk.i.10), gives a reason for that delay of the Parousia which it is one of the main themes of the "Little Apocalypse" to account for, cf. "The End is not yet" (i.7). In Mark's view the End is postponed in order to allow time for the conversion of the Gentiles, which this persecution and its resultant "testimony" will help forward. But Matthew's abbreviation of Mark "As a witness to them and to the Gentiles" (Mt.x.18) misses this point. Again Matthew's

Whosoever shall give to drink
one of these little ones a cup of cold water (Mt.x.42)

 is clearly secondary to Mk.ix.41,

Whosoever shall give you a cup of cold water;

for Matthew's addition "one of these little ones" is derived from another saying of Mark which occurs in the immediate context (Mk.ix.42), so that Matthew is (unconsciously) conflating two passages in Mark. On the other hand, the saying

He that receiveth you receiveth me,
and he that receiveth me receiveth him that sent me
(Mt.x.40, cf. Mk.ix.37)

is quite possibly a case where Mark and M overlap. At any rate it is worthwhile noticing that this saying occurs in four slightly different forms in the Gospels, and is one of those cases (cf. p. 185) where the incorporation by different authors of different versions of a widely circulated quasi-proverbial saying is quite as probable as dependence on a written source.

ὁ δεχόμενος ὑμᾶς ἐμὲ δέχεται, καὶ ὁ ἐμὲ δεχόμενος δέχεται τὸν ἀποστείλαντά με.


ὃς ἂν ἓν τῶν τοιούτων παιδίων δέξηται ἐπὶ τῷ ὀνόματί μου, ἐμὲ δέχεται: καὶ ὃς ἂν ἐμὲ δέχηται, οὐκ ἐμὲ δέχεται ἀλλὰ τὸν ἀποστείλαντά με.


ὁ ἀκούων ὑμῶν ἐμοῦ ἀκούει, καὶ ὁ ἀθετῶν ὑμᾶς ἐμὲ ἀθετεῖ: ὁ δὲ ἐμὲ ἀθετῶν ἀθετεῖ τὸν ἀποστείλαντά με.


ἀμὴν ἀμὴν λέγω ὑμῖν, ὁ λαμβάνων ἀν τινα πέμψω ἐμὲ λαμβάνει, ὁ δὲ ἐμὲ λαμβάνων λαμβάνει τὸν πέμψαντά με.


  1. Chap, i., the Parable chapter, is obviously an agglomeration compiled by the editor of the Gospel. The parable of the Sower, with the narrative introduction and the explanation appended, must be from Mk.iv.1-20. The Mustard and Leaven stood together as a pair in Q. The other four parables are from M or some other source. Thus the evidence of compilation from at least three different sources is conclusive.
  2. The Apocalyptic chapter, Mt.xxiv., is simply Mk.i. ingeniously expanded with material from Q. [This conclusion is not affected if we suppose, with some critics, that Matthew had before him, in addition to Mark's version, a copy of the Little Apocalypse with some slight textual variations.] It is worthwhile for the student, if only on account of the light it throws on Matthew's editorial method of agglomeration, to look up the passages, and see how ingeniously they are suited to their context in Mark. Mt.xxiv.26-28 and 37-39 are fragments of Lk.xvii.22-37, while 43-51 = Lk..42-46. Lastly, if we examine Mt.x.and xviii. closely, we shall see that many of the sections parallel to Mark look less original in their Matthean than in their Marcan form; and the inferiority is of a kind that suggests the hand of an editor.
  3. Matthew xviii. consists very largely of material peculiar to the First Gospel.
    It contains two items, the Lost Sheep and the saying on Forgiveness (Mt.xviii.12-14, 15, 21-22), which in Luke's version differ so much that it is improbable that both can be derived from Q. Both of these may be provisionally assigned to M. But the context in which Matthew places the discourse, as well as the structure of the first half of it, are determined for him by the context and structure of the discourse in Mark (Mk.ix.33-50). Matthew habitually abbreviates Mark, and xviii.8-9 is clearly a contracted version of Mk.ix.43-48. The offending hand and foot have a verse each in Mark, but Matthew combines it into one sentence, "if thy hand or thy foot offend." Again, in Matthew the two words "eternal fire" (v.8) and the addition "of fire" (v. 9) are a brief substitute for Mark's quotation from Is.Ixvi.24, "where the worm dieth not and the fire is not quenched." A comparison of Mk.ix.42 with both Mt.xviii.6-7 and Lk.xvii.1-2 makes it fairly certain that both Mark and Q must have contained the saying about " offending little ones," but that Q contained it with the addition which appears in Lk.xvii.1 = Mt.xviii.7. But Luke, and therefore probably Q, connected the saying on Offences with that on Forgiveness. It does not, however, follow that the saying on Forgiveness, Mt.xviii., is derived from Q. Matthew knew Q as well as M; he may well have put the similar saying on Forgive­ness from M in the same discourse as that which contains the Q saying on Offences.

We conclude that an analysis of every one of the Great Discourses yields evidence that it is an agglomeration put together by the editor of the Gospel.

the infancy narratives

The phrase "Four Document Hypothesis," I need hardly say, is no more intended than was the older term, "Two Document Hypothesis," to rule out the view that the first two chapters in either Matthew or Luke may have been derived from written sources. That is a subject not strictly relevant to the present chapter; nor is it one on which I feel I have anything of much value to contribute. Lest, however, I should seem to ignore altogether so interesting a section of the Gospels, I will take this opportunity briefly, and without going elaborately into reasons, to state my own conclusions.

For the first two chapters of Matthew I see no reason to postulate a written source. For them, as for the narrative additions of the First Gospel, the local tradition of the Church— probably Antioch—where that Gospel was written seems an adequate source. With Luke the case is otherwise. Professor C. C. Torrey
[The Translations made from the, Original Aramaic Gospels, The Ma
cmillan Company, New York, 1912, pp. 290 ff.] argues on linguistic grounds that Lk.i.-ii. must have been translated, not merely from a Semitic language, but from Hebrew as distinct from Aramaic.
[The point is worked out still further in the article, "The Ten Lucan Hymns of the Nativity in their Original Language," by R. A. Aytoun, J.T.S., July 1917.] The point is one on which I have not the linguistic qualifications needed to pro­nounce a judgement. But on one point I feel fairly clear. The Magnificat and the Benedictus were not originally written in Greek. No one who thought in Greek could have produced, either

ἐποίησεν κράτος ἐν βραχίονι αὐτοῦ i.51, or ἠγειρε κέρας σωτηρίας ἡμῖν ἐν οἰκῳ Δαυὶδ i.69.

The question whether the narrative as a whole, as distinct from the hymns it embodies, was written in Hebrew is more difficult. Linguistically it has been pointed out that Lk.i. and ii. are replete both with words and expressions characteristic of Luke's style and also with reminiscences of the LXX. These two observations, however, to some extent cover the same ground; one of the things by which Luke's style is distinguished from that of the other Gospels is his fondness for Septuagintal language. Luke knew his Greek Bible very well, and may have thought a kind of "biblical Greek" appropriate for a Gospel.

A similar consideration would apply, if we supposed that Luke derived these chapters from a Hebrew source. Whoever wrote them was familiar with, and had modelled his style on, the accounts of the birth and infancy of Samson and Samuel in the Old Testament. But, just as modem archaeologists translate Babylonian documents into a style modelled on that of the Authorised Version of the English Bible, so it would be natural to the translator of a Hebrew Protevangelium to adopt the familiar wording of the LXX. Again, Luke himself, if he had a Greek translation of the document in question, would deal with it in the same way as he does with his other sources; he would slightly abbreviate and polish up the Greek, but in this case his very considerable literary instinct would lead him to do the re­writing in Septuagintal Greek.

Taken all in all, the probabilities point to a written source. A question, then, of special interest arises: Did the document as it came to Luke include any indication of a Virgin Birth? In Matthew the virginity of Mary pervades the whole story; for, as we have seen above (p. 87) the reading, "Joseph ... begat Jesus," Mt.i.16, in Syr. S. has small claim to be regarded as the true text. But in Luke extraordinarily little emphasis is laid on it. Indeed, if, with Syr. S. a b c ff 2, we read γυναικί, instead of μεμνυστευμένη in ii.5, the idea of virginity is only clearly brought out in i.34,

And Mary said unto the angel,
How shall this be, seeing that I know not a man?

It is notable that the Old Latin MS. b omits this verse, substituting for it

And Mary said,
Behold, the handmaid of the Lord;
be it unto me according to thy word.

In the ordinary text these words occur later on, as the first half of verse 38; but b omits them in 38, and in this omission it has the support of e. This partial support of b by e may be accidental, but it makes it harder to brush aside the reading of b as an idiosyncrasy of the scribe of that MS. [By a curious coincidence this same MS. is the only one that preserves the reading in Jn.i.13, quoted by various Western fathers (cf. p. 70 above), which makes John assert the Virgin Birth.]  
And as the reading of
b makes excellent sense, the possibility must be considered that it represents the text as Luke wrote it, the ordinary text being a piece of harmonistic editing intended to make it clear that Luke as well as Matthew attached importance to the Virginity. But the question whether the reading of b should be regarded as original is not one which anyone is likely to decide purely on grounds of textual criticism. Those who believe that Christ was born of a Virgin will think it improbable that Luke should have neglected to make this clear, and will scoff at the idea of rejecting the evidence of all the Greek MSS. and all the versions in favour of that of a single Latin MS. of the fifth century. On the other hand, those who regard the Virgin Birth as improbable, but are aware of the immense importance attached to the belief by the Fathers at least as early as Ignatius of Antioch, AD 115, will think it remarkable that a reading which ignores it should have survived till so late a date even in a single MS.


By making it possible to connect the sources of our Gospels with the great Churches, a Four Document Hypothesis explains a wider range of phenomena than the Two Document Hypothesis at present current.

  1. It gives a fuller meaning to the reference in the Preface of Luke to the "many" who had written previously and to the plan and purpose of his own work. Luke knew that several little Churches had their own collection, larger or smaller, of sayings of Christ and stories about Him; but nobody could be sure where they came from or how far they could be trusted.  Luke therefore brushes them all aside. He will use only the materials collected by himself in Caesarea or those of which the authenticity is attested by their reception in the great Churches of Antioch and Rome. The "accuracy" of these materials he can guarantee to Theophilus by reason of his own connection, and the connection of the tradition of these Churches, with the names of those who had been from the beginning "eye-witnesses" like Peter or "ministers of the word" like Philip or like Mark.
  2. It also explains the curious mixture in Matthew of Judaistic with Universalistic sayings, and the concurrence of conspicuously ancient along with some highly doubtful matter. Luke's Gospel bears the impress of an individuality, Matthew has more of an official quality; there is less literary freedom, more careful conflation of written sources. This, too, is explained if we think of Matthew as a studiedly conservative combination of the "gospels" of the three Churches whose traditions would seem to carry the greatest weight, Jerusalem, Antioch, and Rome, expanded with an account of the Infancy and some details of the Passion derived from oral tradition current in the author's own Church most probably the Church of Antioch.
  3. The connection of these Gospels with the traditions of the great Churches explains the authoritative position, which, as against all rivals, they so soon achieved, and thus their ultimate selection as the nucleus of the Canon. It was because the Synoptic Gospels included what each of the great Churches most valued in its own local traditions, and much more also, that the records of these local traditions were allowed to perish.

Thus a Four Document Hypothesis not only offers an extremely simple explanation of all the difficulties which the Two Document Hypothesis cannot satisfactorily meet, but also reflects far better the historical situation in the primitive Church. But there is one thing it does not do. It does not enable us to make a "tidy" scheme showing us exactly which sayings or incidents belong to M, which to L, and which to Q. If Matthew and Luke used four sources, every one of which to a certain extent overlapped with every other, the problem of disentangling them, beyond a certain point, is one that no amount of ingenuity can solve.

But so far as the historian is concerned, this is a matter of very little importance. Only if Q is regarded as the earliest and most authentic of all sources, is it of any special interest to know whether or not it included a particular saying. There was a time when a special authority was attributed to anything which occurred "in all Four Gospels"; again, "the triple tradition" sounded impressive till it was pointed out that a statement of Mark did not become more certain because it happened to be copied by both Matthew and Luke; it is now seen that even the "double tradition" has no special sanctity. So far as historical detail is concerned, Mark and Luke are more to be relied on than Matthew; and where Mark and Luke conflict, Mark is more often to be followed. But as regards the teaching of Christ, much that occurs in a single Gospel is as likely to be genuine as what occurs in two or in all three.

But there is still a value in a "double attestation." If a saying occurs in Q we know for certain that it was written down at a date considerably earlier than that at which the existing Gospels of Matthew and Luke were composed probably also earlier than Mark. Of a saying that is not in Q, all we can say is that this may have been the case. Whenever, however, we find a saying or parable occurring in two different versions whether it be in Q and Mark, Q and M, Q and L, M and L, or M and Mark we have evidence that the saying in question has come down by two different lines of tradition, which probably bifurcated at a date earlier even than that at which Q was written down.

Thus the final result of the critical analysis, which has led to our formulating the Four Document Hypothesis, is very materially to broaden the basis of evidence for the authentic teaching of Christ.